static void Main()
{
string folder = @"c:\temp\slips";
List <string> results = new List <string>();
List <SlipData> allSlips = new List <SlipData>();
foreach (var filename in Directory.GetFiles(folder, "*.pdf"))
{
try
{
List <string> content = PDFWrapper.ReadFile(filename);
if (content.Count > 40)
{
SlipData slipData = new SlipData(content);
if (slipData.IssuesPresent())
{
Console.WriteLine($"{slipData.PayDate} : {slipData.Issues[0]}");
}
allSlips.Add(slipData);
results.Add(JsonConvert.SerializeObject(slipData));
}
}
catch (Exception)
{
// ---- ignore throw;
}
}
foreach (var xx in allSlips)
{
if (xx.Year() == 2018)
{
Console.WriteLine($"{xx.PayDate} {xx.TaxPeriodNumber} {xx.NetPay}");
}
}
// that old pauser!
_ = Console.ReadLine();
}
public static List <string> ReadFile(string filename)
{
List <PDFWrapper> fragments = new List <PDFWrapper>();
try
{
//get text fragments out of file into <fragments>
using (var pdfFile = PdfReader.Open(filename, PdfDocumentOpenMode.ReadOnly))
{
foreach (var page in pdfFile.Pages)
{
PDFWrapper fragment = new PDFWrapper();
var zz = ContentReader.ReadContent(page);
foreach (var element in zz)
{
if (element is COperator op)
{
if (op.Name == "Td")
{
fragment = new PDFWrapper();
fragment.AddPos(op);
}
else if (op.Name == "Tj")
{
fragment.AddText(op);
fragments.Add(fragment);
}
}
}
}
}
}
catch (Exception e)
{
throw e;
}
//write the fragments into the retval
var retval = new List <string>();
double prevY = 0.0;
string str = "";
foreach (var f in from frag in fragments
orderby frag.Y descending, frag.X ascending
select frag)
{
if (prevY != f.Y && str != "")
{
retval.Add(str);
str = "";
}
str += f.ToString();
prevY = f.Y;
}
if (str != "")
{
retval.Add(str);
}
return(retval);
}